Identification of Residues in the C-terminal Domain of HIV-1 Integrase That Mediate Binding to the Transportin-SR2 Protein

Background: The molecular details of the TRN-SR2/HIV-1 IN interaction are not known.

Results: Crucial amino acids in IN for the interaction with TRN-SR2 are located in the CTD, Arg-262/Arg-263/Lys-264, and Lys-266/Arg-269.

Conclusion: TRN-SR2 primarily interacts with the CTD domain of IN.

Significance: Understanding of the IN/TRN-SR2 interaction is necessary to guide drug discovery efforts targeting the nuclear entry step of replication.

Abstract

Transportin-SR2 (TRN-SR2 and TNPO3) is a cellular cofactor of HIV replication that has been implicated in the nuclear import of HIV. TRN-SR2 was originally identified in a yeast two-hybrid screen as an interaction partner of HIV integrase (IN) and in two independent siRNA screens as a cofactor of viral replication. We have now studied the interaction of TRN-SR2 and HIV IN in molecular detail and identified the TRN-SR2 interacting regions of IN. A weak interaction with the catalytic core domain (CCD) and a strong interaction with the C-terminal domain (CTD) of IN were detected. By dissecting the catalytic core domain (CCD) of IN into short structural fragments, we identified a peptide (INIP₁, amino acids ¹⁷⁰EHLKTAVQMAVFIHNFKRKGGI¹⁹¹) retaining the ability to interact with TRN-SR2. By dissecting the C-terminal domain (CTD) of IN, we could identify two interacting peptides (amino acids ²¹⁴QKQITKIQNFRVYYR²²⁸ and ²⁶²RRKVKIIRDYGK²⁷³) that come together in the CTD tertiary structure to form an exposed antiparallel β-sheet. Through site-specific mutagenesis, we defined the following sets of amino acids in IN as important for the interaction with TRN-SR2: Phe-185/Lys-186/Arg-187/Lys-188 in the CCD and Arg-262/Arg-263/Lys-264 and Lys-266/Arg-269 in the CTD. An HIV-1 strain carrying K266A/R269A in IN was replication-defective due to a block in reverse transcription, confounding the study of nuclear import. Insight into the IN/TRN-SR2 interaction interface is necessary to guide drug discovery efforts targeting the nuclear entry step of replication.

Introduction

The human immunodeficiency virus type 1 (HIV-1) belongs to the family of Lentivirinae, characterized by the ability to infect both dividing and nondividing cells (1, 2). After virus entry and reverse transcription, the preintegration complex (PIC)⁵ translocates to the nucleus, and the viral DNA is integrated into the host chromatin. HIV crosses the nuclear membrane through active nuclear import of the PIC (3). As efficient integration into human chromatin depends on nuclear import, and only few PICs seem to enter the nucleus, this process constitutes a true bottleneck for viral replication (4). Until recently, the mechanism of HIV nuclear import, albeit intensely studied, remained obscure (for recent reviews see Refs. 5, 6). Several viral elements, including matrix, viral protein R, capsid, IN, and the central polypurine tract have been suggested to play a role in the nuclear import of HIV (5, 7–10). Next to different viral components, multiple nuclear import factors have been implicated in the nuclear translocation of the PIC, such as importin-α/β (11–13), importin-7 (14–16), and importin-α3 (17). Recently, transportin-SR2 (TRN-SR2, TNPO3) has been proposed as a cellular cofactor of HIV-1 replication (18–20) mediating nuclear import of the PIC (19, 20). Independent of two genome-wide siRNA screens (18, 20), TRN-SR2 was identified as an interaction partner of HIV-1 integrase in a yeast two-hybrid screen (19). A reverse screen confirmed the interaction and demonstrated that none of the other HIV proteins directly interact with TRN-SR2. Using RNAi technology, the crucial role of TRN-SR2 in HIV replication was demonstrated (18–20). A specific reduction in two long terminal repeat (LTR) circles and integration was measured by quantitative PCR (Q-PCR) (19, 21, 22). Direct evidence for TRN-SR2-mediated nuclear import of HIV was obtained by using fluorescently labeled PICs (19). The importance of TRN-SR2 for HIV replication has been confirmed by multiple reports, but its exact mechanism of action is still under debate (7, 18–28). HIV-1 integrase (IN) mediates the integration of the viral DNA into the host chromatin (29). It consists of three domains connected by flexible linkers as follows: 1) the N-terminal HH-CC zinc-binding domain (NTD); 2) the catalytic core domain (CCD), and 3) the C-terminal domain (CTD) (30). Integrase has been implied in nuclear import of HIV in multiple studies, and various nuclear localization sequences have been reported. First, a bipartite nuclear localization signal in the CTD was described that consisted of amino acids ²¹¹KELQKQITK²¹⁹ and ²⁶¹PRRKAK²⁶⁶ (11). Later, a nonclassical NLS ¹⁶¹IIGQVRDQAEHLK¹⁷³ was described in the CCD (31). Although TRN-SR2 was identified in a yeast two-hybrid screen as an interaction partner of HIV-1 IN (19), the question still remains whether the interaction between IN and TRN-SR2 mediates HIV nuclear import. As a first step in answering this question, we analyzed the interaction of TRN-SR2 and HIV-IN in molecular detail to identify the minimal interacting regions and the protein-protein interaction hot spots in HIV IN.

EXPERIMENTAL PROCEDURES

Recombinant Protein Purification

Recombinant proteins were expressed in Escherichia coli BL21-CodonPlus (DE3). Recombinant His₆-tagged HIV-1 integrase was purified as described previously (32). We thank Dr. Woan-Yuh Tarn (Institute. of Biomedical Sciences, Taiwan) for the pGEX-TRN-SR2 expression plasmid. Recombinant GST-tagged and His-tagged TRN-SR2 were purified as described previously (19). For the expression of the GST peptides, bacteria were grown to an OD of 0.6, and protein expression was induced with 0.5 mm isopropyl β-d-thiogalactopyranoside. After incubation at 37 °C for 2 h, the bacteria were harvested, washed, and stored at −20 °C. For purification of the GST peptides, the cells were resuspended in lysis buffer (PBS (pH 7.4), 0.5 m NaCl, 1 mm DTT, 1 mg/ml lysozyme, 0.1 mm PMSF, 1 μl of DNase/10 ml). After complete lysis by sonication, the supernatant was cleared by centrifugation, and recombinant proteins were bound to glutathione-Sepharose resin (GE Healthcare). After washing of the resin with 20 volumes of washing buffer (PBS (pH 7.4), 0.5 m NaCl, 1 mm DTT), the GST-tagged protein was eluted with 10 ml of elution buffer (PBS (pH 7.4), 0.5 m NaCl, 1 mm DTT, 20 mm reduced glutathione). The fractions were analyzed by SDS-PAGE for protein content, pooled, and dialyzed (overnight, 4 °C) against storage buffer (PBS (pH 7.4), 1 mm DTT, 10% (v/v) glycerol).

AlphaScreen Binding Assay

The AlphaScreen binding assay was optimized for use in 384-well OptiPlate microplates (PerkinElmer Life Sciences) with a final volume of 25 μl. Proteins were all diluted to 5× working solutions in the assay buffer (25 mm Tris (pH 7.4), 150 mm NaCl, 1 mm MgCl₂, 2 mm DTT, 0.1% (v/v) Tween 20, and 0.1% (w/v) bovine serum albumin (BSA)). First, 10 μl of the TRN-SR2 was pipetted into the wells, followed by 5 μl of His₆-IN or a GST-peptide dilution series. The plate was sealed and left to incubate for 1 h at 4 °C. Next, 10 μl of a mixture of Ni²⁺ chelate acceptor and glutathione donor AlphaScreen beads (PerkinElmer Life Sciences) was added. This establishes final concentrations of 20 μg/ml for each of the beads. Plates were then incubated for 1 h at 30 °C and analyzed using an EnVision Multilabel Reader (PerkinElmer Life Sciences) according to the manufacturer's instructions. Each titration was performed in duplicate, and assays were repeated at least twice in independent experiments. The equilibrium dissociation constants (apparent K_d) were calculated with Prism 5 (GraphPad), by plotting the titration curves using a one-site-specific binding with Hill slope fit as shown in Equation 1,

The background signal was subtracted from each value. Points for which quenching of the signal occurs through excess of either binding partner (“hooking” as referred to by the supplier), were excluded from the final plot.

Co-immunoprecipitation

Proteins were expressed by double polyethyleneimine-mediated transfection of 293T cells with 10 μg of each plasmid. 24 h after transfection, cells were harvested and lysed in co-IP buffer (50 mm Tris (pH 7.3), 300 mm NaCl, 1 mm MgCl₂, 10% glycerol (v/v), 0.5% Triton X-100 (v/v) and protease inhibitor mixture (EDTA-free)) after which lysates were incubated with anti-FLAG-agarose beads (Sigma). After 3 h of incubation, the agarose beads were recovered by centrifugation for 2 min at 4000 rpm at 4 °C and washed three times with 500 μl of co-IP buffer. Bound proteins were analyzed by SDS-PAGE followed by Western blotting. Membranes were probed with monoclonal antibodies against 3×FLAG (1:10,000 dilution) (Sigma) and GFP (1:5000) (Abcam). Detection was performed using horseradish peroxidase-conjugated goat anti-mouse and rabbit anti-goat antibodies, respectively (Dako), and chemiluminescence (ECL+, Amersham Biosciences).

Integrase Activity Assays

The 3′-processing activity was measured in a standard assay using radioactively labeled oligodeoxynucleotide substrates resembling the viral LTR end, although a quantitative ELISA was used to measure the overall enzymatic activity (33).

Viruses and Viral Vectors

The viral molecular clones pNL4-3 and pNL4-3.Luc.R-E− were obtained through the AIDS Research and Reference Reagent Program. The pMD.G plasmid encodes the vesicular stomatitis virus glycoprotein (34, 35). Viruses were produced by single polyethyleneimine-mediated transfection of 293T cells using 20 μg of the viral molecular clone pNL4-3 per 10-cm dish. Vesicular stomatitis virus glycoprotein-pseudotyped viral vectors were produced by double polyethyleneimine-mediated transfection of 293T cells using 5 μg of the envelope expression plasmid pMD.G and 20 μg of the molecular clone pNL4-3.Luc.R-E− per 10-cm dish. Two and 3 days post-transfection, the supernatant was harvested, filtered with 0.22-μm pore-size syringe filters (Sartorius), concentrated by centrifugal filtration using VivaSpin concentrators (Millipore), and treated with DNase I (Roche Applied Science).

RESULTS

Determination of the Interaction Domain between HIV-1 Integrase and TRN-SR2

By yeast two-hybrid, we previously identified TRN-SR2 as a bona fide binding partner of HIV-1 IN (19). A reverse screen confirmed the interaction between TRN-SR2 and IN and demonstrated that no other viral protein interacts with TRN-SR2 under these conditions. By now, the interaction has independently been confirmed by co-IP, pulldown (7, 19), AlphaScreen (26), and surface plasmon resonance (7).

To define the minimal TRN-SR2 interaction domain in HIV-1 integrase, we now investigated its interaction with the NTD, the CCD, and the CTD. The different IN domains fused to GFP were expressed in 293T cells, and TRN-SR2 was expressed with a 3×FLAG tag (Fig. 1A). By co-IP, TRN-SR2 was shown to interact with both the catalytic core and the C-terminal domain of IN but not with the NTD (Fig. 1A). Next, deletion mutants of either the CCD or the CTD were generated and tested for interaction with TRN-SR2 in co-IP (data not shown). This approach failed to generate further insights because mutants bound nonspecifically to the agarose beads, most likely due to their reduced structural stability. We therefore opted for an in vitro approach.

FIGURE 1.

TRN-SR2 interacts with the catalytic core domain and with the C-terminal domain of IN. A, GFP-IN and FLAG-TRN-SR2 were detected with anti-GFP and anti-3×FLAG antibodies, respectively, after Western blotting. GFP-tagged full-length IN or IN domains were co-expressed with 3×FLAG-TRN-SR2 or 3×FLAG in 293T cells (left panel). 3×FLAG-TRN-SR2 and 3×FLAG were immunoprecipitated (IP) with anti-FLAG antibody, in the presence of full-length IN, ΔNTD (IN deleted for NTD), ΔCTD (IN deleted for CTD), NTD, CCD, or CTD of IN (right panel). 3×FLAG serves as a control for aspecific co-immunoprecipitation. B, analysis of the interaction of GST-tagged TRN-SR2 with His₆-tagged full-length integrase (upper left panel), the NTD (upper right panel), the CCD (lower left panel), and the CTD (lower right panel) of HIV-IN. Complexes of GST-TRN-SR2 and His₆-IN were bound to glutathione donor beads and nickel-chelate acceptor beads. Light emission was measured using an EnVision Multilabel Reader. Each experiment was repeated three times. Representative graphs of single experiments are shown. The apparent equilibrium dissociation constants (K_d) were calculated with GraphPad Prism 5 and are indicated.

We purified recombinant full-length IN and its domains carrying an N-terminal His₆ tag as well as recombinant TRN-SR2 with an N-terminal GST tag to determine the interaction by AlphaScreen (Fig. 1B). The apparent K_d value as well as the AlphaScreen counts are important to compare the affinity of two proteins tested in AlphaScreen. Both the CCD and the CTD of HIV-1 IN displayed affinity for TRN-SR2. The CCD of IN interacts with TRN-SR2 with a 2-fold lower affinity (K_d = 40.2 ± 15.1 nm) and 10-fold lower counts than full-length IN (K_d = 16.8 ± 3.8 nm), whereas the CTD of IN interacts with TRN-SR2 in a cooperative manner with a 6-fold lower affinity (K_d = 94.2 ± 2.6 nm) but 4-fold higher counts than full-length IN (Fig. 1B). The data confirm direct protein-protein interactions between TRN-SR2 and both CCD and CTD of HIV-1 IN as shown by co-IP. Hence, we focused our subsequent analysis of the interaction interface on these two domains.

Defining the Minimal Interaction Domain in the CCD of HIV-1 IN

Nuclear localization signals are usually composed of short amino acid stretches and are commonly bipartite (36). Classical NLSs for instance are known to be composed of one or two basic-rich stretches (37). Therefore, single point mutations usually do not abolish the interaction of a cargo with its specific nuclear import factor (38, and multiple mutations are required. Likewise, abolishing the IN/TRN-SR2 interaction by single point mutations proved to be difficult in our hands, and deletion mutants of IN domains displayed poor solubility.⁶ Therefore, we opted for a different strategy, selecting short oligopeptide stretches that are lining the surface of the CCD of HIV-1 IN and screened those for interaction with TRN-SR2. For ease of expression and purification, we produced the small CCD fragments with a GST tag and analyzed their binding to His₆-TRN-SR2. Reversal of the affinity tags did not affect the interaction of IN with TRN-SR2 (data not shown; GST-IN/His₆-TRN-SR2, K_d = 18.6 ± 1.9 nm; His₆-IN/GST-TRN-SR2, K_d = 16.8 ± 3.8 nm). We purified four distinct GST-tagged oligopeptide fragments representing the four α-helices lining the CCD surface (Table 1) and analyzed them for their affinity to His₆-TRN-SR2 (Fig. 2A).

TABLE 1.

Amino acid sequences of IN derived fragments

Fragment name	Fragment sequence
CCD(91–108)a	AETGQETAYFLLKLAGRW
CCD(118–138)	GSNFTSTTVKAACWWAGIKQE
CCD(145–166)	QSQGVVESMNKELKKIIGQVRD
CCD(170–191)	EHLKTAVQMAVFIHNFKRKGGI
CTD(214–228)	QKQITKIQNFRVYYR
CTD(229–243)	DSRDPLWKGPAKLLW
CTD(244–258)	KGEGAVVIQDNSDIK
CTD(259–273)	VVPRRKVKIIRDYGK
CTD(274–288)	QMAGDDCVASRQDED

^a The fragment numbers refer to the positions of the amino acids in integrase and define the first and the last amino acid of each CTD fragment. The peptides were designed to match protein stretches (15–22 amino acids) of IN with at least some surface-exposed amino acids.

FIGURE 2.

Critical amino acids at the surface of the INIP₁ in the CCD of HIV-1 IN. A, analysis of the interaction between TRN-SR2 and the different CCD peptides. The fragment numbers refer to the positions of the amino acids in integrase and define the first and the last amino acid of each CCD peptide. ●, CCD(91–108); ■, CCD(118–138); ▴, CCD(145–167); ▾, CCD(170–191)_. B, monomer of the catalytic core of IN is shown (PDB code 1EX4 (36)). The aromatic amino acid Phe-185 is colored green, the acidic amino acid Glu-170 is red, and basic amino acids His-171/Lys-173/Lys-186/Arg-187/Lys-188 are blue. AlphaScreen analysis of the affinity of the different mutant INIP₁ (C) and mutant CCDs (D) for TRN-SR2 is given. ●, wild type; ■, Glu-170/His-171/Lys-173; ▴, Phe-185/Lys-186/Arg-187/Lys-188. Complexes were bound to glutathione donor beads and nickel-chelate acceptor beads. The signal, light emission, was measured using an EnVision Multilabel Reader (in counts/s). Each experiment was repeated three times. Representative graphs of single experiments are shown.

One of the four peptides (CCD(170–191)) displayed a moderate affinity for His₆-TRN-SR2 (Fig. 2A). The AlphaScreen signal for CCD(170–191) allowed us to determine a K_d of 401.4 ± 7.7 nm. We will refer to the CCD(170–191) peptide as integrase-derived interaction peptide 1 (INIP₁). To map the amino acids involved in the interaction, we performed mutagenesis of selected residues in INIP₁. The selection of mutants was guided by the general knowledge of NLS sequences and the surface exposure of amino acid residues in integrase (Fig. 2B). Therefore, we generated two mutants of INIP₁ (INIP₁^{E170Q/H171A/K173A} and INIP₁^{F185A/K186A/R187A/K188A}) and assayed their affinity for TRN-SR2 (Fig. 2C). Both mutants of INIP₁ displayed a reduced affinity for TRN-SR2. To confirm that these amino acids were the main contributors for the IN-TRN-SR2 interaction, we generated the corresponding mutants in the CCD domain (CCD^{E170Q/H171A/K173A} and CCD^{F185A/K186A/R187A/K188A}). In agreement with the results obtained with the peptides, CCD^{E170Q/H171A/K173A} showed a reduced affinity for TRN-SR2, although CCD^{F185A/K186A/R187A/K188A} did not interact any longer with TRN-SR2 in AlphaScreen (Fig. 2D).

Defining the Minimal Interaction Domain in the C-terminal Domain of HIV-1 IN

Following the same approach, we divided the C-terminal domain of HIV-1 IN into short oligopeptide stretches of 15 amino acids each (Table 1) and screened those for interaction with TRN-SR2. The small CTD fragments were purified with a GST tag. The five distinct GST-tagged oligopeptide fragments together covered amino acids 214–288 of the CTD and were analyzed for their affinity to His₆-TRN-SR2 in the AlphaScreen assay (Fig. 3A).

FIGURE 3.

CTD of integrase contains two interaction hot spots for TRN-SR2 binding that can be combined to a high affinity binding peptide. A, analysis of the interaction between TRN-SR2 and the different CTD peptides. ▴, CTD(214–229); ♦, CTD(229–244); ●, CTD(244–259); ■, CTD(259–274); ▾, CTD(274–288). B, left, structure of the CCD-CTD integrase (PDB code 1EX4 (36)). Amino acids 214–229 (CTD(214–229)) are colored in dark green. Amino acids 259–274 (CTD(259–274)) are colored in light green. Right, putative structure of the fragment encompassing IN amino acids from 214 to 229 and from 262 to 274 linked with a three-amino acid linker. C, AlphaScreen analysis of the affinity of CTD(214–229)-(link-262–274) for His₆-TRN-SR2 in comparison with the CTD(214–229) and CTD(259–274). ▴, CTD(214–229); ■, CTD(259–274); ♦, CTD(214–229)-(link-262–274). The fragment numbers refer to the positions of the amino acids in integrase and define the first and the last amino acid of each CTD fragment. Complexes were bound to glutathione donor beads and nickel-chelate acceptor beads. The signal, light emission, was measured using an EnVision Multilabel Reader (in counts/s). Each experiment was repeated three times. Representative graphs of single experiments are shown.

Two out of five peptides (CTD(214–229) and CTD(259–274)) displayed affinity for His₆-TRN-SR2. CTD(214–229) interacts with TRN-SR2 with an apparent K_d of 288.3 ± 22.03 nm, whereas the interaction with CTD(259–274) was too weak to allow accurate K_d determination (Fig. 3A). Mapping the sequences of both peptides onto the crystal structure of CCD-CTD integrase demonstrates a striking structural relationship between both peptides; CTD(214–229) and CTD(259–274) together form an antiparallel β-sheet in the CTD of IN (Fig. 3B).

We subsequently coupled fragments CTD(214–229) and CTD(259–274) through a flexible three-amino acid linker (DSG). Based on structural information (PDB code 1EX4 (39)) amino acids 259–261 were left out, resulting in the 30-mer CTD(214–228)-(link-262–273) (Fig. 3B). The affinities of CTD(214–228), CTD(259–273), and the combined peptide were subsequently determined (Fig. 3C). The combined peptide CTD(214–228)-(link-262–273) interacts with TRN-SR2 with a similar affinity (K_d = 87.9 ± 4.8 nm) as the affinity of the entire CTD (K_d = 108.5 ± 3.6 nm) (Fig. 3C). The combined peptide will be referred to as integrase-derived interaction peptide 2 (INIP₂).

Mapping of the Hot Spots of Interaction in INIP₂

To pinpoint the actual protein-protein contacts, we performed mutagenesis of selected residues in INIP₂. Selection of mutants was again based on the general knowledge of NLS sequences and amino acid surface exposure in IN (Fig. 4A). Because a typical NLS contains a set of basic residues, we first generated mutant INIP₂ versions in which two or three basic residues were substituted with alanine. Some NLSs, like the M9 NLS (37), are rich in glycine and aromatic residues. Hence, we also mutated several aromatic residues into alanine. We then tested these mutant peptides for interaction with TRN-SR2 in the AlphaScreen assay (Fig. 4B). Substitutions of either positively charged or aromatic amino acids, affected the binding of INIP₂ to TRN-SR2. INIP₂^R224A/R228A (K_d = 173.8 ± 10.9 nm) displayed a slightly lower affinity for TRN-SR2 than wild type INIP₂ (K_d = 113.0 ± 0.6 nm). INIP₂^K266A/R269A (K_d = 298.5 ± 28.8 nm) and INIP₂^{R262A/R263A/K264A} (K_d = 384.1 ± 23.7 nm) displayed a 3–4-fold lower affinity than wild type INIP₂, respectively (Fig. 4B). The binding of INIP₂^F223A/Y226A was too weak to allow confident K_d determination.

FIGURE 4.

Critical amino acids at the surface of the INIP₂ in the C-terminal domain of HIV-1 IN. A, aromatic amino acids Phe-223/Tyr-226 are colored green, and basic amino acids Arg-224/Arg-228/Arg-262/Arg-263/Lys-264/Lys-266/Arg-269 are blue. The structure is based on PDB code 1EX4 (36). B, AlphaScreen analysis of the affinity for TRN-SR2 of the different mutant INIP₂. ●, wild type; ▴, Arg-224/Arg-228; ■, Phe-223/Tyr-226; ▾, Arg-262/Arg-263/Lys-264; ♦, Lys-266/Arg-269. C, AlphaScreen analysis of the affinity for TRN-SR2 of different mutant CTDs. ●, wild type; ▾, Phe-223/Arg-224/Tyr-226/Arg-228; , Arg-262/Arg-263/Lys-264; ♦, Lys-266/Arg-269. Complexes were bound to glutathione donor beads and nickel-chelate acceptor beads. The signal, light emission, was measured using an EnVision Multilabel Reader (in counts/s). Each experiment was repeated three times. Representative graphs of single experiments are shown.

As none of the initial substitutions led to a complete loss of binding, we generated two CTD mutants each containing two sets of substitutions generated in INIP₂ (CTD^{F223A/R224A/Y226A/R228A} and CTD^{R262A/R263A/K264A/K266A/R269A}). CTD^{F223A/R224A/Y226A/R228A} showed a reduced affinity for TRN-SR2, although CTD^{R262A/R263A/K264A/K266A/R269A} completely lost binding to TRN-SR2 in AlphaScreen (Fig. 4C). To narrow down the number of substitutions needed to abrogate the binding of the CTD to TRN-SR2, we generated two more CTD mutants (CTD^{R262A/R263A/K264A} and CTD^K266A/R269A). CTD ^K266A/R269A showed a reduced affinity for TRN-SR2 although CTD^{R262A/R263A/K264A} completely lost binding to TRN-SR2 (Fig. 4C).

IN Mutants Defective for Interaction with TRN-SR2

To prove the importance of the selected residues in the CCD and CTD, we finally inserted these sets of substitutions into full-length IN. Interaction of both IN^{F185A/K186A/R187A/K188A} and IN^{R262A/R263A/K264A} with TRN-SR2 was reduced in comparison with WT IN (Fig. 5A). The F185A/K186A/R187A/K188A substitutions only moderately affected TRN binding (50% reduction of AlphaScreen signal), whereas the R262A/R263A/K264A substitutions clearly mediated the largest effect (95% reduction in signal), although residual binding subsists (Fig. 5A). To control for correct folding of the interaction-defective mutant integrases (IN^{F185A/K186A/R187A/K188A} and IN^{R262A/R263A/K264A}), we analyzed their interaction with LEDGF/p75 (32) and IN (Fig. 5B). LEDGF/p75 is a validated cofactor of integrase that interacts with integrase through its CCD domain (40–42). The development of a robust AlphaScreen to measure interaction between LEDGF/p75 and IN has been described before (33). IN^{R262A/R263A/K264A} bound LEDGF/p75 with a similar affinity as wild type IN (Fig. 5B) demonstrating that the overall structure of IN is not altered by these substitutions. Not surprisingly, IN^{F185A/K186A/R187A/K188A} did not bind to LEDGF/p75 (Fig. 5B) because the substitutions K186A/R187A are known to inhibit tetramerization and LEDGF/p75 binding (43). To additionally control for the tertiary and quaternary structure, we tested the mutant integrases by size exclusion chromatography (Fig. 5C). The F185A/K186A/R187A/K188A substitutions clearly affected the IN oligomerization as no tetramers could be measured, only dimers of IN where present. The R262A/R263A/K264A substitutions, however, did not affect IN oligomerization (Fig. 5C). Because the F185A/K186A/R187A/K188A substitutions clearly affect IN multimerization (Fig. 5C), we cannot exclude that the impact on TRN-SR2 binding is due to the altered multimerization equilibrium. Therefore and because the F185A/K186A/R187A/K188A substitutions (located in the CCD) only modestly affected TRN-SR2 binding, we focused on the substitutions in C-terminal IN for the remainder of our study.

FIGURE 5.

Analysis of IN^{F185A/K186A/R187A/K188A} and IN^{R262A/R263A/K264A} with reduced affinity for TRN-SR2. A, analysis of the interaction between IN^{F185A/K186A/R187A/K188A} and IN^{R262A/R263A/K264A} and TRN-SR2. B, analysis of the interaction between IN^{F185A/K186A/R187A/K188A} and IN^{R262A/R263A/K264A} and wild type LEDGF/p75. ●, wild type IN; ▴, IN^{F185A/K186A/R187A/K188A}; ▾, IN^{R262A/R263A/K264A}. Complexes were bound to glutathione or anti-FLAG donor beads (for GST-TRN-SR2 and LEDGF/p75, respectively) and nickel-chelate acceptor beads. The signal, light emission, was measured using an EnVision Multilabel Reader (in counts/s). The experiment was repeated three times. Representative graphs of single experiments are shown. C, size exclusion chromatography elution profiles of WT IN, IN^{F185A/K186A/R187A/K188A}, and IN^{R262A/R263A/K264A}. 15 μg of each IN was loaded onto a Superose 6 10/300 GL column connected to an Äkta FPLC. Peaks corresponding to tetrameric (Tet) IN or dimeric (Dim) IN are indicated.

Analysis of the Replication of HIV-1 Defective for Interaction with TRN-SR2

Because IN substitutions are known to have pleiotropic effects on HIV replication, we tried to find a set of mutations that only affects interaction with TRN-SR2. We screened different sets of double point mutants for their effect on binding to TRN-SR2 (Fig. 6A). Ultimately, we selected two double mutant INs (IN^R263A/K264A and IN^K266A/R269A) with a 10-fold reduced affinity for TRN-SR2. We next assayed these two mutants for 3′-processing and overall integrase activity. IN^R263A/K264A was defective (<25% of WT activity) for 3′-processing and strand transfer, whereas IN^K266A/R269A could catalyze 3′-processing (>75% of WT activity) but not strand transfer (data not shown). As productive DNA binding is a prerequisite of catalytic activity, this assay indirectly reflects on the DNA binding capacity of IN^K266A/R269A. Because IN^K266A/R269A showed a strongly reduced affinity for TRN-SR2 and retained 3′-processing activity, we selected this mutant for further virological studies. To study the impact of the loss of interaction between IN^K266A/R269A and TRN-SR2 on the nuclear import of HIV, the IN^K266A/R269A mutations were inserted into a molecular clone of HIV (pNL4-3). Mutant and wild type viruses were harvested after transient transfection of 293T cells and normalized both on p24 levels and RT activity, which showed comparable ratios. HeLaP4 cells were infected in parallel with WT or IN^K266A/R269A virus, and supernatants were collected at days 3–7 post-infection. The IN^K266A/R269A virus proved to be replication-deficient (Fig. 6B). To pinpoint at which step the viral replication was inhibited, we analyzed a single round HIV-1 infection using Q-PCR. Late reverse transcripts were measured by Q-PCR at different time points after infection (44, 45). Reduced levels of late reverse transcripts were evidenced for NL4-3-IN^K266A/R269A virus compared with WT NL4-3 (Fig. 6C). We also produced a vesicular stomatitis virus glycoprotein-pseudotyped IN^K266A/R269A NL4-3.Luc.R-E− vector and compared it with a WT pseudotyped NL4-3.Luc.R-E− vector using Q-PCR (Fig. 6D). Again only background levels of reverse transcription were evidenced confounding the study of nuclear import.

FIGURE 6.

IN^K266A/R269A virus is replication-deficient due to a reverse transcription block. A, comparison of the interaction between mutant integrases and TRN-SR2. ●, wild type IN; ▴, IN^R262A/R263A; ♦, IN^R262A/K264A; ■, IN^R263A/K264A; ▾, IN^K264A/K266A; , IN^K266A/R269. Complexes were bound to glutathione donor beads and nickel-chelate acceptor beads. The signal, light emission, was measured using an EnVision Multilabel Reader (in counts/s). The experiment was repeated three times. Representative graphs of single experiments are shown. B, HeLaP4 cells were infected with 5 × 10⁴ pg of p24 of HIV NL4-3 WT or IN^K266A/R269A mutant virus. From 2 days postinfection, supernatants were sampled daily for p24 measurements. One of two independent experiments each performed in duplicate is shown. ●, HIV NL4-3 WT; ■, HIV NL4-3 IN^K266A/R269A. C, HeLaP4 cells were infected with 10 × 10⁶ pg of p24 HIV NL4-3 WT or IN^K266A/R269A mutant. At distinct time points after infection, DNA was extracted from cells, and viral DNA species were quantified by Q-PCR. ●, HIV NL4-3 WT with AZT; ■, HIV NL4-3 WT with DMSO; ▴, HIV NL4-3 IN^K266A/R269A with AZT; ▾, HIV NL4-3 IN^K266A/R269A with DMSO. D, HeLaP4 cells were infected with 10 × 10⁶ pg of p24 HIV NL4-3.Luc.R-E− WT or IN^K266A/R269A mutant vector. At distinct time points after infection, DNA was extracted from cells, and viral DNA species were identified by Q-PCR. ●, HIV NL4-3.Luc.R-E− WT with AZT; ■, HIV NL4-3.Luc.R-E− WT with DMSO; ▴, HIV NL4-3.Luc.R-E− IN^K266A/R269A with AZT; ▾, HIV NL4-3.Luc.R-E− IN^K266A/R269A with DMSO. AZT, azidothymidine.

DISCUSSION

Recently, multiple research groups have reported on the importance of TRN-SR2 as a cellular cofactor of HIV replication (7, 18–28). Although the exact role of TRN-SR2 in HIV replication is still under debate, there is a consensus that even moderate knockdown of this cofactor results in a strong inhibition of HIV replication at an early replication step prior to integration. Because TRN-SR2 belongs to the importin-β family of nuclear import factors (46–49), it is not far-fetched to imagine a direct or indirect role in the nuclear import of the viral PIC (7, 19, 21). Recently, the hypothesis was formulated that TRN-SR2 also acts as a nuclear export factor for tRNA and HIV capsid (27), but this model awaits independent confirmation.

We identified TRN-SR2 as an interaction partner of HIV-1 IN in a yeast two-hybrid screen (19). Reverse yeast two-hybrid screening with TRN-SR2 only revealed IN as a viral binding partner. By employing Q-PCR and imaging of fluorescent viral particles, we could demonstrate that knockdown of TRN-SR2 blocks HIV replication at the step of nuclear import, a bottleneck for the infection of the host cell (19). Whether the physical interaction of TRN-SR2 with HIV-1 IN is the sole determinant for the nuclear import of the PIC and whether TRN-SR2 exerts other roles prior to and/or after nuclear import of the PIC still await further investigation (7, 18–28). Here, we studied the molecular interaction between TRN-SR2 and HIV-1 IN to define the NLS in IN. Because IN has been proposed before to play a role in HIV nuclear import (5, 6, 11, 12, 50), several attempts have been made to define potential NLSs in IN. A bipartite NLS composed of two stretches ranging from amino acids ²¹¹KELQKQITK²¹⁹ and ²⁶¹PRRKAK²⁶⁶ (11) and a nonclassical NLS composed of amino acids ¹⁶¹IIGQVRDQAEHLK¹⁷³ were described (31). Multiple IN mutations were inserted into viruses and tested for their impact on viral replication. IN mutants affecting HIV nuclear import were assayed for reduced a two-LTR circle formation (51, 52). Despite these extensive studies, so far no single NLS was unambiguously identified as the determinant of HIV nuclear import. These older studies however were hampered by the lack of knowledge on the identity of the nuclear import factor.

Extended NLS sequences that result in the recognition of structural motifs in import factors are commonly found in cargo. To ensure a certain redundancy in cargo binding, it is unlikely that a single point mutation in the cargo will abrogate the interaction. With this in mind, we searched for the interaction domain in IN by assaying IN peptide stretches rather than using scanning or deletion mutagenesis. Mapping of the IN/TRN-SR2 interface was based on dissecting a large interaction domain into several small peptides and assaying those for their affinity for TRN-SR2. This method solved problems encountered with protein stability and solubility and allowed us to define INIP₁ and INIP₂. Of note, the GST-peptide approach can be recommended for other protein-protein interaction studies when protein deletions are unstable or multipartite interaction regions occur.

Co-immunoprecipitation experiments demonstrated that TRN-SR2 interacts with both the catalytic core and the C-terminal domain of IN. Interaction studies with recombinant proteins confirmed the lack of firm contact with the NTD but revealed strong binding between TRN-SR2 and the CTD and weaker binding with the CCD prompting us to focus on these domains for further analysis.

In the original TRN-SR2 yeast two-hybrid paper, we described a virtual selected interaction domain based on the overlap between multiple clones that were picked up (19). The identified selected interaction domain covered the CCD, but a lot of the clones that where picked up also contained the CTD (19). As such, the identified clones further corroborate the present identification of a major interaction spot in the CTD and a weak second interaction spot in the CCD.

Here, we identified one stretch of amino acids in the CCD (¹⁷⁰EHLKTAVQMAVFIHNFKRKGGI¹⁹¹) defined as the first integrase-derived interaction peptide (INIP₁) and two stretches of amino acids (²¹⁴QKQITKIQNFRVYYR²²⁸ and ²⁵⁹VVPRRKVKIIRDYGK²⁷³) in the CTD. Of note, the peptides identified contain the potential NLS sequences that where previously identified (11, 31) but not confirmed due to a lack of an established interaction partner. By combining the two stretches of the CTD, we defined the second integrase-derived interaction peptide (INIP₂). By mutagenesis of INIP₁ and INIP₂, we identified two stretches of amino acids that are crucial for the interaction. Phe-185/Lys-186/Arg-187/Lys-188 determine interaction in the CCD although Arg-262/ Arg-263/Lys-264 and Lys-266/Arg-269 are hot spots in the CTD. In the accompanying paper by Larue et al. (53), the role of the CTD residues Arg-262, Arg-263, and Lys-264 in binding to TRN-SR2 was independently confirmed.

After confirming their role in the CCD and the CTD, we combined the mutations in full-length IN and generated an IN mutant IN^{F185A/K186A/R187A/K188A-R262A/R263A/K264A} that does not bind TRN-SR2. Because the F185A/K186A/R187A/K188A substitutions affect IN multimerization (Fig. 5C), we cannot exclude that the impact on TRN-SR2 binding is due to a modified multimerization equilibrium. Therefore and because the CCD substitutions only modestly affected TRN-SR2 binding, we focused on substitutions in the CTD for the remainder of our study. Because IN substitutions are known to have pleiotropic effects on HIV replication, we tried to find a set of mutations affecting only the interaction with TRN-SR2. We screened different sets of double point mutations for their binding to TRN-SR2. We selected two mutants (IN^R263A/K264A and IN^K266A/R269A) with a strongly reduced affinity for TRN-SR2. Next, we assayed these two mutants for 3′-processing and overall integrase activity. Because IN^K266A/R269A showed 10-fold reduced affinity for TRN-SR2 and retained 3′-processing activity, we selected this mutant for further virological studies. We introduced the mutations in a molecular clone and studied HIV replication. The K266A/R269A substitutions in IN rendered HIV replication-defective. However, due to a reverse transcription defect, we could not assess the impact of loss of IN/TRN-SR2 interaction on the nuclear import. The pleiotropic effects of IN mutations, especially in the CTD, are well known (51). First, the region between amino acids 220 and 270 of HIV IN plays a role in DNA binding (54). Furthermore, IN is a key protein within the viral particle that is not only responsible for integration catalysis but also interacts with other viral and cellular proteins to ensure integrity of the PIC. Within these nucleoprotein complexes, IN is known to interact with the reverse transcriptase. The interface of this interaction has been mapped to three crucial amino acids (K258A, W243E, and V250E) (55), but IN mutations outside of this interface (e.g. K186Q, R228A, R262A/R263A, R262A/K264A, K266A, and R269A) were shown to affect reverse transcription as well (51, 52, 56). Unfortunately, all mutants identified in this study are known to display reduced reverse transcription during replication (51, 52, 56), confounding the study of their role in the nuclear import of HIV in cellular replication assays. Still, insight into the hot spots that define the protein-protein interaction between IN and TRN-SR2 is necessary to guide future drug discovery efforts targeting the nuclear entry step of replication.